Update README.md

README.md

@@ -25,8 +25,9 @@ trained on `166` protein sequences in the [RNA binding sites dataset](https://hu
 using a `85/15` train/test split. This model was trained with class weighting due to the imbalanced nature 
 of the RNA binding site dataset (fewer binding sites than non-binding sites). This model has slightly improved 
 precision, recall, and F1 score over [AmelieSchreiber/esm2_t12_35M_weighted_lora_rna_binding](https://huggingface.co/AmelieSchreiber/esm2_t12_35M_weighted_lora_rna_binding)
-but may suffer from mild overfitting, as indicated by the training loss being slightly lower than the eval loss. If you are searching for 
-binding sites and aren't worried about false positives, the higher recall may make this model preferable to the other RNA binding site predictors. 
+but may suffer from mild overfitting, as indicated by the training loss being slightly lower than the eval loss (see metrics below). 
+If you are searching for binding sites and aren't worried about false positives, the higher recall may make this model 
+preferable to the other RNA binding site predictors. 
 
 You can train your own version 
 using [this notebook](https://huggingface.co/AmelieSchreiber/esm2_t6_8M_weighted_lora_rna_binding/blob/main/LoRA_binding_sites_no_sweeps_v2.ipynb)!